AUSTRALASIAN JOURNAL OF COMBINATORICS
Volume 53 (2012), Pages 273–284
On 2-dominating kernels in graphs
Andrzej Wloch
Rzeszów University of Technology
Faculty of Mathematics and Applied Physics
al. Powstańców Warszawy 12, 35-359 Rzeszów
Poland
awloch@prz.edu.pl
Abstract
In this paper we introduce new kinds of kernels in graphs. Using existing
concepts of an independent set and a 2-dominating set, we define in a
natural way the concept of 2-dominating kernels in graphs. We characterize some classes of graphs having a 2-dominating kernel, also using the
idea of a local 2-dominating kernel.
1
Introduction
In general we use standard terminology and notation of graph theory; see [5], [12].
Consider a finite, connected graph G with vertex set V (G) and edge set E(G). The
cardinality of V (G) is the order of G and the cardinality of E(G) is its size. By
Pn we mean a graph with the vertex set V (Pn ) = {t1 , . . . , tn } and the edge set
E(Pn ) = {ti ti+1 : i = 1, . . . , n − 1}, n ≥ 2. Moreover P1 is the graph that contains
only one vertex. If xy ∈ E(G), then we say that x is a neighbour of y. The set
of all neighbours of x is called the open neighbourhood of x and it is denoted by
N(x). The set N(x) ∪ {x} is the closed neighbourhood and it is denoted by N[x].
For a subset X ⊂ V (G) we write N(X) and N[X] instead of
N(x) and
N[x],
x∈X
x∈X
respectively. By dG (x, y) we denote the distance between vertices x and y in G. For
a vertex x ∈ V (G) let degG (x) denote its degree.
A graph H is a subgraph of a graph G if V (H) ⊆ V (G) and E(H) ⊆ E(G). If
every edge of E(G) with both end-vertices in V (H) is in E(H), we say that H is
induced by X ⊂ V (G), and we write H = G[X] or H ≤ G if the set X is known. If
H is a subgraph of G, then we say that G is a supergraph of H. A vertex x ∈ V (G)
is a simplicial vertex if N[x] induces a complete subgraph. Then N[x] is named as
a simplex of G. If every vertex of G belongs to a simplex, then G is a simplicial
graph. In a simplicial graph a vertex which is not simplicial will be named as a
nonsimplicial vertex. Let H be an arbitrary graph. For a graph G of order n, n ≥ 3,
H-addition (or adH ) stands for a local augmentation of G which is the operation
274
ANDRZEJ WlOCH
G → adH(x,y)(G) of adding a graph H to the graph G by identifying the vertex x in
G with the vertex y in H.
A vertex v of a graph G is called a cut vertex of G if G − v has more components
than G.
Recall that a vertex of degree 1 is called a pendant vertex (or a leaf). For
x ∈ V (G), denote by L(x) the set of leaves attached to the vertex x. A vertex
x ∈ V (G) such that L(x) = ∅ is a support vertex. The set of support vertices in G
we denote by S(G). Moreover if |L(x)| ≥ 2, then x is named as a strong support
vertex. If L(x) = 1, then x is a weak support vertex and the unique pendant vertex
adjacent to a weak support vertex will be called a single leaf. The set of pendant
vertices in a graph G attached to the strong support vertices (respectively, weak
support vertices) in G we denote by Ls (G) (respectively, Lw (G)) and consequently
L(G) = Ls (G) ∪ Lw (G) is the set of all pendant vertices in G.
A subset S ⊆ V (G) is an independent set of G if no two vertices of S are
adjacent in G. Moreover, a subset containing only one vertex and the empty set also
are independent. An independent set S is maximal if there is no independent set of
G containing S as a proper subset.
A subset Q ⊆ V (G) is a dominating set of G if each vertex from V (G) \ Q
has a neighbour in Q. A subset J ⊂ V (G) which is independent and dominating
simultaneously is a kernel.
The literature includes many papers dealing with the theory of independent sets
and kernels in graphs.
The classical concept of kernels in a digraph was introduced in [13] by Neumann
and Morgenstern in the context of the game theory as independent and dominating
sets. Since then the concept has been relevant in graph theory for its relations
with a variety of problems: in list colourings and perfectness. Berge was one of the
pioneers in this area studying the existence of kernels in digraphs and successfully
using kernels to solve problems in others areas of mathematics; see [2].
In the literature there are some variants and generalizations of kernels, for example (k, l)-kernels which generalize kernels in distance sense or kernels by monochromatic paths which generalize kernels with respect to edge-colouring of a graph. Recently there are some interesting results concerning (k, l)-kernels and kernels by
monochromatic paths in digraphs; see for instance papers [6, 7, 8, 9, 10, 11] and
[15], [17, 18].
In an undirected graph every maximal independent set is a kernel. The problem
is more complicated if we add additional restrictions, studying for example special
kernels named in the literature as efficient dominating sets, see [1]. In this paper we
study kernels in this direction.
A subset Q ⊆ V (G) is a 2-dominating set of G if each vertex from V (G) \ Q has
at least two neighbours in Q. The concept of 2-dominating sets has been studied in
the literature; see for example [3, 4]. In this paper we formally define a new kind of
kernel in graphs using the concept of independence and 2-dominating sets of G.
2-DOMINATING KERNELS IN GRAPHS
275
A subset J ⊂ V (G) is a 2-dominating kernel of G if J is independent and 2dominating. Clearly a 2-dominating kernel of G is a kernel of G. From the definition
of J it follows that if G is connected, then G has order at least 3 and the set J
has at least two vertices. Note that if G is totally disconnected, then V (G) is a
2-dominating kernel of G.
Moreover the definition of the set J implies that every pendant vertex of a graph
G belongs to a 2-dominating kernel. The definition of the set J gives the following
observation.
Fact 1.1 Let L(G) ⊂ V (G) be the set of pendant vertices of a graph G. If J is a
2-dominating kernel of G, then J is a maximal independent set of G such that L ⊆ J.
Fact 1.2 Let G be a graph of order n, n ≥ 5. If G has a 2-dominating kernel, then
G \ N[Ls (G)] has a 2-dominating kernel.
Maximal independent sets including the set of pendant vertices were introduced
by Wloch in [16] and next they were studied in special graphs such as trees and
unicyclic graphs, in particular with respect to the number NMIL (G) defined as the
number of maximal independent sets including the set of pendant vertices as a subset.
For trees the following has been proved.
Theorem 1.3 [16] Let T be an n-vertex tree with n ≥ 3. Then NMIL (T ) ≤ P v(n −
3), where P v(n) is the Padovan number defined by P v(0) = P v(1) = P v(2) = 1 and
P v(n) = P v(n − 2) + P v(n − 3), n ≥ 3.
Evidently every graph has at least one maximal independent set including the
set of pendant vertices as a subset, but an arbitrary graph G does not always have a
2-dominating kernel; for example, it is easy to see that a graph P4 does not possess
a 2-dominating kernel.
In this paper we give a characterization of special classes of graphs and their
products with a 2-dominating kernel.
2
Graphs with 2-dominating kernels
We first consider paths and cycles and we establish necessary and sufficient conditions
for the existence of 2-dominating kernels in these graphs. These results follow by
simple observation, so we omit the proofs.
Theorem 2.1 A graph Pn has a 2-dominating kernel J if and only if n = 2p + 1,
p ≥ 1 and this kernel is unique. If it holds, then |J| = p + 1.
Theorem 2.2 A graph Cn has a 2-dominating kernel J if and only if n = 2p, p ≥ 2
and there are exactly two disjoint 2-dominating kernels. If it holds, then |J| = p.
276
ANDRZEJ WlOCH
Theorem 2.3 If G is a graph of order n, n ≥ 3 in which every vertex is pendant or
it is a strong support vertex, then G has a 2-dominating kernel J = L(G).
Theorem 2.4 Let G = G(V1 , V2 ) be a bipartite graph. If for each two pendant
vertices x, y ∈ V (G), dG (x, y) ≡ 0 (mod 2), then G has a 2-dominating kernel.
Proof: Assume that G = G(V1 , V2 ) is a bipartite graph. Let L(G) be the set of
pendant vertices. Because for each two pendant vertices x, y ∈ V (G), dG (x, y) ≡
0 (mod 2), so L(G) ⊆ V1 . Moreover every vertex u ∈ V2 has degree at least two, or
otherwise it would be in V1 . This immediately gives that J = V1 is a 2-dominating
kernel of a bipartite graph. Thus the theorem is proved.
2
Next we give some necessary conditions for graphs having 2-dominating kernels.
Theorem 2.5 Let G be a graph of order n, n ≥ 3 and size m, m ≥ 2. If G has a
2-dominating kernel J, then |J| ≥ n − m2 . Moreover the equality holds if and only if
G is a bipartite graph G = G(V1 , V2 ) with V1 = J and V2 = V (G) \ J such that for
each x ∈ V2 , degG x = 2.
Proof: Let J be a 2-dominating kernel of a graph G. Because J is independent, we
have that the set J induces the edgeless subgraph on |J| vertices. Moreover every
vertex v ∈ V (G) \ J has at least two edges incident to the set J, so degG v ≥ 2.
Because we have n − |J| vertices outside the set J, so 2(n − |J|) ≤ m. Then by
simple calculations the result follows.
To prove the equality it suffices to observe that if in a bipartite graph G, for every
vertex v ∈ V (G) \ J we have degG v = 2, then m = 2(n − |J|), which concludes the
necessity.
Assume now that |J| = n − m2 . We shall show that G is bipartite with V1 = J
and V2 = V (G) \ J and for each x ∈ V2 , degG x = 2. Clearly V1 = J. It suffices to
prove that V (G) \ J induces an edgeless subgraph and for each x ∈ V2 , degG x = 2.
Because |V2 | = |V (G) \ J| = n − |J| = m2 , hence m = 2|V2 |. From the assumption J
is 2-dominating hence for each x ∈ V2 , degG x ≥ 2 and by m = 2|V2 | it immediately
follows that degG x = 2, for each x ∈ V2 . This means that no two vertices from V2
are joined by an edge. Consequently V2 induces the edgeless graph, so G = G(V1 , V2 )
is a bipartite graph. Thus the theorem is proved.
2
Theorem 2.6 Let T be an arbitrary tree on n vertices, n ≥ 3. If T has a 2. Moreover the equality holds for T P2k+1 ,
dominating kernel J, then |J| ≥ n+1
2
k ≥ 1.
Proof: If n = 3, then the theorem is obvious. Let n ≥ 4. Let J be a 2-dominating
kernel of T and suppose that J = {u1 , . . . , up }, where p ≥ 2. Then there are n − p
vertices x1 , . . . , xn−p ∈ V (T ) \ J. Since J is a 2-dominating set of T , then without
2-DOMINATING KERNELS IN GRAPHS
277
loss of generality assume that x1 u1 ∈ E(T ) and x1 u2 ∈ E(T ), where u1 = u2 . Then
for x2 ∈
/ J there is at least one vertex from the set J different from u1 , u2 ∈ J, say u3 ,
such that u3 dominates x2 . Otherwise if u1 and u2 dominate x2 then there is a cycle
u1 − x2 − u2 − x1 − u1 in T , a contradiction that T is a tree. Proving analogously
for the remaining vertices from the set V (T ) \ J we deduce that for n − p vertices
from V (T ) \ J we need at least n − p + 1 vertices in the set J. Otherwise T contains
a cycle, a contradiction that T is a tree. Consequently n ≥ n − p + n − p + 1 and by
simple calculations we obtain that |J| = p ≥ n+1
. The equality immediately follows.
2
Thus the theorem is proved.
2
Theorem 2.7 Let x be a weak-support vertex of G and u be the single leaf adjacent
to x. If G has a 2-dominating kernel, then (N(x) \ {u}) \ S(G) = ∅.
Proof: Let J be a 2-dominating kernel of a graph G. Clearly a support vertex
cannot be in a 2-dominating kernel because every pendant vertex has to belong to a
2-dominating kernel.
Let x be a weak-support vertex and assume on the contrary that (N(x) \ {u}) \
S(G) = ∅. This means that all neighbours (except the vertex u) of x are support
vertices. Consequently for each v ∈ (N(x) \ {u}), v ∈
/ J. This means that x has
exactly one neighbour in J, a contradiction that G has a 2-dominating kernel. Thus
the theorem is proved.
2
Theorem 2.8 Let G be a graph of order n, n ≥ 3. Suppose that G contains two
subgraphs isomorphic to Pt and Pm with t, m ≥ 2 such that
(i) Pt and Pm have an endpoint x in common, and
(ii) the endpoints of Pt and Pm distinct from x are leaves in G.
If G has a 2-dominating kernel, then t and m are either both odd or both even.
Proof: Let J be a 2-dominating kernel of a graph G and assume on the contrary
that t is odd and m is even. Let x be the common vertex of Pt and Pm . Clearly the
endpoints of Pt and Pm distinct from x belong to J. Since t is odd, then by Theorem
2.1 also x ∈ J, otherwise the vertex u ∈ N(x) ∩ V (Pt ) has exactly one neighbour in
J, a contradiction to the fact that a graph G has a 2-dominating kernel. Because
m is even the vertex v ∈ N(x) ∩ V (Pm ) belongs to J, or otherwise its neighbour y,
y = x, has only one neighbour in J. But then v, x ∈ J and moreover vx ∈ E(G),
contradicting the independence of J. Thus the theorem is proved.
2
The following theorems characterize graphs with 2-dominating kernels in terms
of their special subgraphs.
Theorem 2.9 Let G be a graph of order n, n ≥ 5. Suppose that G contains p,
p ≥ 2, subgraphs isomorphic to Pt1 ,. . . ,Ptp with ti ≥ 2, i = 1, . . . , p such that
278
ANDRZEJ WlOCH
(i) Pt1 ,. . . ,Ptp have only a vertex x in common, and
(ii) the endpoints of Pti , i = 1, . . . , p distinct from x are leaves in G, and
(iii) degG x = p + 1.
Let either ti , i = 1, . . . , p be odd or ti , i = 1, . . . , p be even. For all ti , i = 1, . . . , p
odd, a graph G has a 2-dominating kernel if and only if G \ (
p
i=1
Pti \ {x}) has a
2-dominating kernel. For all ti , i = 1, . . . , p even, a graph G has a 2-dominating
kernel if and only if G \
p
i=1
Pti has a 2-dominating kernel.
Proof: Assume that G is as in the statement of the theorem and consider the
following possibilities:
(1). ti is odd for all i = 1, . . . , p.
Let J be a 2-dominating kernel of G. Since L(G) ⊆ J so all endpoints of Pti ,
i = 1, . . . , p, distinct from x, belong to J. Let J ∩ V (Pti ) = Ji , i = 1, . . . , p. Then
it is clear that J = J ∪
p
i=1
Ji , where J = J ∩ V (G \
p
i=1
Pti ). The definition of
2-dominating kernel and ti odd imply that x ∈ Ji , for each i = 1, . . . , p. Because
degG x = p + 1 in the graph G \ (
p
i=1
∗
Pti \ {x}), the vertex x has to belong to the 2-
dominating kernel J . Consequently we deduce that J ∗ = J ∪{x}. Assume now that
G\
p
i=1
Pti has a 2-dominating kernel J . Let G = G \
p
i=1
Pti . Since degG (x) = p + 1,
hence degG x = 1 and x ∈ J . Let Pti be an ti vertex path, i = 1, . . . , p with pendant
vertex yi . Let Ji be a 2-dominating kernel of the subgraph Pti , i = 1, . . . , p. Clearly
yi ∈ Ji , for all i = 1, . . . , p. Adding graphs Pti , i = 1, . . . , p to the graph G by
identifying the vertex x in G with the vertex yi in Pti we augment a graph G to a
graph G such that J ∪
p
i=1
Ji is a 2-dominating kernel of G.
(2). ti is even for all i = 1, . . . , p.
Proving analogously as for the previous case, we obtain that N(x) ∩
and vertex x ∈
/ J is dominated by p, p ≥ 2, vertices from N(x) ∩
in G = G \
p
i=1
Pti the set J ∗ = J \ (
p
i=1
p
i=1
p
i=1
V (Pti ) ⊂ J
V (Pti ) . Hence
V (Pti )) is a 2-dominating kernel of G .
Similarly as in case (1), using a local augmentation of a graph G , we construct a
2-dominating kernel in a graph G. Thus the theorem is proved.
2
Using the same method as above we can prove.
Theorem 2.10 Let G be a graph of order n, n ≥ 5. Suppose that G contains a
subgraph Pt , t ≥ 4, such that the endpoint y of Pt is a leaf in G. Let Pt−2
be a
subgraph of Pt including the leaf y. If G has a 2-dominating kernel, then G \ Pt−2
has a 2-dominating kernel.
2-DOMINATING KERNELS IN GRAPHS
279
The theorems above show that we can reduce the starting graph G to smaller a
graph and instead of G we can study this subgraph with respect to the existence of
a 2-dominating kernel.
Theorem 2.11 A simplicial graph G has a 2-dominating kernel if and only if every
simplex of G has exactly one simplicial vertex and every nonsimplicial vertex belongs
to at least two simplices.
Proof: Let J be a 2-dominating kernel of a simplicial graph G. Let M1 , . . . , Mk
be simplices of G. Clearly V (G) =
k
i=1
V (Mi ). Since every simplex is a complete
subgraph, so |J ∩V (Mi )| ≤ 1, for all i = 1, . . . , k. We shall show that |J ∩V (Mi )| = 1,
for each i = 1, . . . , k. Assume on the contrary that there is a simplex Mi , 1 ≤ i ≤ k,
such that J ∩ V (Mi ) = ∅. This means that for each x ∈ V (Mi ), N(x) ⊆ Mi and we
obtain a contradiction that Mi has a simplicial vertex. Consequently |J ∩V (Mi )| = 1.
We claim that every simplex Mi , i = 1, . . . , k has exactly one simplicial vertex.
Suppose on the contrary that there is 1 ≤ i ≤ k such that Mi has two simplicial
vertices u, v ∈ V (Mi ). Because J is a 2-dominating set, we deduce that u, v ∈ J, a
contradiction to |J ∩ V (Mi )| = 1. Since J is 2-dominating, for each x ∈ V (Mi ), if
x is a nonsimplicial vertex then |N(x) ∩ J| ≥ 2. From the above considerations we
find that every nonsimplicial vertex has to belong to at least two simplices.
Assume now that every simplex of G has exactly one simplicial vertex and every
nonsimplicial vertex belongs to at least two simplices. Let mi ∈ Mi be a simplicial
vertex, for i = 1, . . . , k. We shall show that J ∗ =
k
i=1
{mi } is a 2-dominating kernel
of G. Since mi ∈ Mi , i = 1, . . . , k is a simplicial vertex, for each simplicial vertex
mj ∈ Mj , i = j we have that mj ∈ N[mi ]. Consequently mi mj ∈ E(G) and
this immediately gives that J ∗ is independent. Let x ∈ V (G) \ J ∗ . Then x is a
nonsimplicial vertex and by our assumption there are simplices Mt , Mr , t = r, such
that x ∈ Mt and x ∈ Mr . Moreover, there are simplicial vertices mt ∈ Mt and
mr ∈ Mr such that mt , mr ∈ J ∗ . All this gives xmt ∈ E(G) and xmr ∈ E(G), so J ∗
is a 2-dominating set of G. Thus the theorem is proved.
2
Now in studying 2-dominating kernels we use the concept of a local 2-dominating
kernel which is a useful technical tool in Kernel Theory. A local 2-dominating kernel
J ⊆ V (G) is an independent set that is locally 2-dominating, i.e. N(J) ⊆ V (G) \ J
and every vertex from N(J) has at least two neighbours in J. It is worth mentioning
that a 2-dominating kernel is a local 2-dominating kernel, while a local 2-dominating
kernel J satisfying the condition N[J] = V (G) is a 2-dominating kernel. A feasible
solution for a graph G is a sequence of induced subgraphs and local 2-dominating
kernels < Gi , Ji >i=1,...,n such that
(1) G1 = G;
(2) Ji is a local 2-dominating kernel of Gi for all 1 ≤ i ≤ n − 1;
(3) Gi+1 = Gi \ N[Ji ] for all 1 ≤ i ≤ n − 1;
(4) Jn is a 2-dominating kernel of Gn .
280
ANDRZEJ WlOCH
Theorem 2.12 A graph G has a 2-dominating kernel if and only if G has a feasible
solution.
Proof: It is obvious that if J is a 2-dominating kernel, then G1 , J1 is a feasible
solution of G.
Assume now that Gi , Ji i=1,...,n is a feasible solution of graph G. We shall show
that J =
n
i=1
Jn is a 2-dominating kernel of G. To prove independence of the set J,
assume, for a contradiction, that there are x, y ∈ J such that xy ∈ E(G). Because
every local 2-dominating kernel is independent and the set J is defined by union of
local 2-dominating kernels, we assume that x ∈ Ji and y ∈ Jj , for i = j. Without
lose of generality assume that i < j. Then y ∈ N[Ji ], since Ji is a local 2-dominating
kernel. Then by the definition of a feasible solution y ∈
/ V (Gj ) and consequently
y∈
/ Jj , a contradiction to the assumption.
Now we prove that J is 2-dominating. Let x ∈ (G \ N[J]) and assume on the
contrary that x does not have at least two neighbours in J. This means that x ∈
/ N[Ji ]
for all i = 1, . . . , n. In this case x ∈ V (Gn ) \ Jn and by the condition (4) the set Jn is
2-dominating of Gn , so it follows that x has two neighbours in Jn , a contradiction to
n
Ji also is 2-dominating. Thus the theorem is proved.
the assumption; hence J =
i=1
2
Now we use the concept of local 2-dominating kernels to give sufficient conditions
for the existence of a 2-dominating kernel in trees.
For an arbitrary tree T we define the sequence η = Ti , Ji i=1,...,n in a graph T
T1 = T
J1 = Ls (T1 )
T2 = T \ N[J1 ]
J2 = Ls (T2 ) and for 3 ≤ i ≤ k
Ti = Ti−1 \ N[Ji−1 ] for all i < n such that Tn does not possess strong support
vertices.
Using the result for feasible solution from Theorem 2.12 we can characterize trees
with 2-dominating kernels using the concept of local 2-dominating kernels.
Theorem 2.13 Let T be an arbitrary tree of order n, n ≥ 3 and let η = Ti , Ji i=1,...,n
be the sequence in T . If for each component T ∗ of Tn ,
(i) T ∗ contains only isolated vertices, or
(ii) for each x, y ∈ L(T ∗ ) we have dT ∗ (x, y) ≡ 0 (mod 2),
then T has a 2-dominating kernel.
Proof: Using Theorem 2.12 it suffices to show that the sequence η = Ti , Ji 1≤i≤n
is a feasible solution for a graph T . By the definition of the sequence η it suffices
to show that the subtree Tn has a 2-dominating kernel. It suffices to consider only
2-DOMINATING KERNELS IN GRAPHS
281
connected components of Tn . Let T ∗ be a connected component of Tn . If T ∗ has an
isolated vertex, say x, then x ∈ J, where J is a 2-dominating kernel of T ∗ . Assume
that for each two pendant vertices in a component T ∗ , we have dT ∗ (x, y) ≡ 0 (mod 2).
Because T ∗ is a bipartite graph so by Theorem 2.4 the result immediately follows.
Thus the theorem is proved.
2
Theorem 2.14 Let x be a cut vertex of G and let G1 , . . . , Gk , k ≥ 2 be connected
components of a graph G\{x}. If for each i = 1, . . . , k a subset V (Gi )∪{x} induces a
subgraph Ri with 2-dominating kernel Ji such that x ∈ Ji , then G has a 2-dominating
kernel.
Proof: Let x be a cut vertex of G and G1 , . . . , Gk , k ≥ 2 be as in the statement of
the theorem. We shall show that J =
k
i=1
Ji is a 2-dominating kernel of G. By the
assumption we have x ∈ J. So it is obvious that J is a 2-dominating set because of
2-dominating sets Ji , i = 1, . . . , k. Moreover, |Ji | ≥ 2, for all i = 1, . . . , k. To prove
independence of J assume that u, v ∈ J. Let u ∈ Jp ⊂ V (Rp ), for some p, 1 ≤ p ≤ k.
If v = x, then either v ∈ V (Rp ) or v ∈ V (Rj ), j = p. If v ∈ V (Rp ), then uv ∈
/ E(G)
by independence of Jp . If v ∈ V (Rj ), j = p, then uv ∈
/ E(G), otherwise we obtain a
contradiction that x is a cut vertex of a graph G. Thus the theorem is proved. 2
Now we describe a local augmentation which preserves the existence of 2-dominating kernels in a graphs.
Theorem 2.15 Let G be an n-vertex graph (n ≥ 3) having a 2-dominating kernel.
Then for an arbitrary m = n + p, p ≥ 1, there is a supergraph R of G of order m
having a 2-dominating kernel.
Proof: Let G be an n-vertex graph, n ≥ 3, with a 2-dominating kernel J. For
an arbitrary p ≥ 1 we will construct a supergraph R of order n + p, p ≥ 1, with
2-dominating kernel J such that J ⊂ J . Let x be an arbitrary vertex from V (G)\J.
We consider two possibilities:
1. x ∈
/J
Applying the local augmentation for graph H = K1,p , p ≥ 1 by identifying vertices
x ∈ V (G) and the centre of the of the star K1,p we obtain the supergraph R ⊃ G
with order n + p. Moreover the set J = J ∪ L(K1,p ) is a 2-dominating kernel of R.
2. x ∈ J
Then it is clear that there is y ∈
/ J such that xy ∈ E(G). Proving, as in case 1, by
replacing x by y, the result follows. Thus the theorem is proved.
2
282
3
ANDRZEJ WlOCH
2-dominating kernels in σ(α, G)
In this section we consider σ(α, G) of graphs with respect to the existence of 2dominating kernels in graphs. To give necessary and sufficient conditions for graphs
it is worth working with graphs whose structure can be described using smaller
graphs. Clearly such results come from the study of products of graphs. Different
operations of graphs allow us to build various families of graphs with 2-dominating
kernels.
In this section we present necessary and sufficient conditions for the existence of 2dominating kernels in G-join σ(α, G). Let G be a graph with V (G) = {x1 , . . . , xn },
n ≥ 2, and α = (Gi )i=1,...,n be a sequence of vertex disjoint graphs on V (Gi ) =
{y1i , . . . , ypi i }, pi ≥ 1, i = 1, . . . , n.
The G-join of the graph G and the sequence α is a graph σ(α, G) such that V (σ(α, G))
=
n
i=1
({xi } × V (Gi )) and E(σ(α, G)) = {(xs , yjs )(xq , ytq ); s = q and yjs yts ∈ E(Gs ) or
xs xq ∈ E(G)}. By Gci we mean a copy of the graph Gi in σ(α, G). The G-join of
graphs is a large-scale graph operation. It is interesting to mention that if V (Gi ) = V
for i = 1, . . . , n, then from G-join we obtain the generalized lexicographic product
G[G1 , . . . , Gn ]. If G1 = G2 = . . . = Gn , then from the G-join we obtain the composition G[H] of two graphs. If G = K2 , then we obtain the join of two graphs
G1 + G2 .
/ I and Gi consists of only
Let X ⊆ V (G) and X = {xi ; i ∈ I}. If Gj = K1 for j ∈
two isolated vertices, i.e. |V (Gi )| = 2 and E(Gi ) = ∅ for i ∈ I, then σ(α, G) gives
the duplication of the subset X ⊆ V (G). In particular if |I| = 1, then we obtain the
definition of the duplication of the vertex xi ∈ V (G).
To prove the main result of this section, firstly we give the independence lemma.
Lemma 3.1 [14] A subset S ∗ ⊂ V (σ(α, G)) is an independent set of σ(α, G) if
and only if S ⊂ V (G) is an independent set of G such that S ∗ =
Si where
i∈I
I = {i : xi ∈ S}, Si ⊂ V (Gci ) is an independent set of Gci , for every i ∈ I.
Theorem 3.2 A graph σ(α, G) has a 2-dominating kernel if and only if there is a
2-dominating kernel J = {xi : i ∈ I}, I ⊂ {1, . . . , n}, of the graph G such that Gi ,
i ∈ I has a 2-dominating kernel.
Proof: Assume that σ(α, G) has a 2-dominating kernel, say J ∗ . We shall prove that
the set J = {xi ∈ V (G) : J ∗ ∩V (Gci ) = ∅} is a 2-dominating kernel of G. It is obvious
by Lemma 3.1 and the definition of σ(α, G) that J ∗ =
Ji , where I ⊂ {1, . . . , n}
i∈I
and I = {i; xi ∈ J}, where J is an independent set of G. It suffices to show that the
set J is 2-dominating. Let xt ∈ (V (G) \ J). Then from the definition of the set J,
for each (xt , yp ), 1 ≤ p ≤ pt , we have (xt , yp ) ∈
/ J ∗ . Since J ∗ is a 2-dominating kernel,
∗
there exist two vertices from J , say (xs , yr ), (xk , yl ) ∈ S ∗ , where s can be equal to k,
such that (xt , yp )(xs , yr ) ∈ E(σ(α, G)) and (xt , yp )(xk , yl ) ∈ E(σ(α, G)). This means
2-DOMINATING KERNELS IN GRAPHS
283
that there are two vertices xs , xk ∈ J such that xt xs ∈ E(G) and xt xk ∈ E(G).
Hence there exists a 2-dominating kernel J of G such that if xi ∈ J, then Gi has a
2-dominating kernel. Let us now suppose that there exists a 2-dominating kernel of
G, say J = {xi : i ∈ I} where I ⊂ {1, . . . , n} such that if xi ∈ J, then Gi has a
2-dominating kernel.
We shall show that σ(α, G) has a 2-dominating kernel. Because Gi has a 2-dominating kernel, so Gci also has a 2-dominating kernel in σ(α, G). By Lemma 3.1 and
the definition of the graph σ(α, G) to obtain a 2-dominating kernel of σ(α, G) we
have to choose a 2-dominating kernel of Gci for each i ∈ I. Evidently the choice of
subset J ∗ =
Ji is a 2-dominating kernel of σ(α, G). Thus the theorem is proved.
i∈I
2
Acknowledgements
The author wishes to thank the referees for a through review and very useful suggestions which improved the rewriting of this paper.
References
[1] D.W. Bange, A.E. Barkauskas and P.J. Slater, Efficient dominating sets in
graphs, Applications of Discrete Math. SIAM, Philadelphia, PA (1988), 189–
199.
[2] C. Berge, Graphs and Hypergraphs, North-Holland, Amsterdam, 1973.
[3] M. Blidia, M. Chellali and O. Favaron, Independence and 2-domination in trees,
Australas. J. Combin. 33 (2005), 317–327.
[4] M. Chellali, Bounds on the 2-domination number in cactus graphs, Opuscula
Mathematica 26 (1) (2006), 5–12.
[5] R. Diestel, Graph theory, Springer-Verlag, Heidelberg, New-York, Inc., 2005.
[6] H. Galeana-Sánchez, Kernels in edge-coloured digraphs, Discrete Math. 184
(1998), 87–99.
[7] H. Galeana-Sánchez and R. Rojas-Monroy, On monochromatic paths and
monochromatic 4-cycles in edge-colored bipartite tournaments, Discrete Math.
285 (2004), 313–318.
[8] H. Galeana-Sánchez and L.A.J. Ramirez, Monochromatic kernel perfectness of
special clases of digraphs, Discuss. Math. Graph Theory 27 (3) (2007), 389–400.
284
ANDRZEJ WlOCH
[9] H. Galeana-Sánchez and R. Gomez, (k, l)-kernels, (k, l)-semikernels, k-Grundy
functions and duality for state splitting, Discuss. Math. Graph Theory 27 (2007),
359–371.
[10] H. Galeana-Sánchez and M. Guevara, Some sufficient conditions for the existence
of kernels in infinite digraphs, Discrete Math. 309 (2009), 3680–3693.
[11] H. Galeana-Sánchez and C. Hernandez-Cruz, On the existence of (k, l)-kernels
in infinite digraphs. A survey (submitted).
[12] T.W. Haynes, S.T. Hedetniemi and P.J. Slater, Fundamentals of Domination in
Graphs, Marcel Dekker, New York, 1998.
[13] J. Von Neumann and O. Morgenstern, Theory of Games and Economic Behavior, Princeton University Press, Princeton 1953.
[14] W. Szumny, A. Wloch and I. Wloch, On (k, l)-kernels in D-join of digraphs,
Discuss. Math. Graph Theory 27 (2007), 457–470.
[15] I. Wloch, On kernels by monochromatic paths in the corona of digraphs, Central
European J. Math. 6 (4) (2008), 537–542.
[16] I. Wloch, Trees with extremal numbers of maximal independent sets including
the set of leaves, Discrete Math. 308 (2008), 4614–4624.
[17] I. Wloch, On imp-sets and kernels by monochromatic paths in duplication, Ars
Combin. 83 (2007), 93–99.
[18] I. Wloch, On kernels by monochromatic paths in D-join, Ars Combin. 98 (2011),
215–224.
(Received 2 Nov 2011; revised 14 Feb 2012)